Mistral releases open weights for trillion-parameter Mistral Large 4
Original title: Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity
Nicknamed Le Chonk, ML4 is on Hugging Face and Mistral's API, and the company claims 47 percent fewer tokens than GLM-5.3.
French company Mistral published weights for Mistral Large 4, a trillion-parameter model introduced earlier this month. Organizations can download them from Hugging Face or call the API and deploy the system themselves, including on premises. Mistral says ML4 matches open-weight model GLM-5.3 while using 47 percent fewer tokens per task and leads other open models on Artificial Analysis' Coding Agent Index. Executives cast the release as sovereign cyber defense that outside providers cannot cut off.
Unconfirmed: Performance versus GLM-5.3 and the Coding Agent Index ranking are Mistral's claims, not independently verified here.
- Weights are on Hugging Face and available through Mistral's API.
- Mistral claims frontier performance at a fraction of closed-lab compute and capital.
- Pierre Stock said open weights cannot be cut off by an outside provider.
- He called rogue-agent risks real but solvable and warned against fearmongering.
- Vendors
- Mistral
- Organizations
- OpenAIAnthropicHugging FaceArtificial Analysis
- AI models
- Mistral Large 4GLM-5.3
From the original article
3 min read at techcrunch.com
Mistral says its new model can afford organizations a kind of control over its cyber defenses that closed models can't.
By Lucas Ropek · techcrunch.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.